[GT] Inconsistencies in date related options for scripts

["Weigert, Thomas" <weigert AT mst.edu>]

I am sure it is no great surprise to you that there are significant
inconsistencies with respect to options for scripts.

I have looked at the date related options in order to see what we can
factor out and get more consistent as a consequence.

Here are my observations:

1. Difference in date related options:

--full (missing in graphic.pl, scan.pl)
--start
--end
--nb-item (missing in backtest.pl, backtest_many.pl, backtest_multi.pl,
display_signal.pl, scan.pl, 
--max-loaded-items (missing in backtest.pl, backtest_many.pl,
backtest_multi.pl, display_indicator.pl, display_signal.pl, scan.pl)

However, in display_indicator.pl, --nb-item is interpreted also as
--max-loaded-items.

2. Difference in treatment of timeframes

Some indicators (graphics.pl, display_indicator.pl, display_signal.pl)
check all available timeframes when loading prices, the others only use
the current timeframe.

3. Differences in algorithm that determines first and last period

graphic.pl uses a different scheme than the other scripts.

4. Differences on the default number of periods considered

Some indicators (graphics.pl) base the default number of periods
analyzed on --nb-item, others use a hardwired constant of 200
(display_indicator.pl, display_signal.pl) independent of timeframe, and
others determine it based on the timeframe ratio (backtest.pl,
backtest_many.pl, backtest_multi.pl, scan.pl).

5. Finally, display_indicator.pl supports --last-record. This would also
make sense for display_signal.pl

In my own installation I have made these consistent long time ago, as I
kept getting confused by when what option is available and when not. I
propose that we make them consistent throughout GT.

Please weigh in with your preferences as to how to resolve conflicts
between how options are handled. There will have to be some changes in
the default number of periods analyzed, but these should be minor.

Please advise.

Th.